A couple of comments:

None of the FPGA synthesis tools cost in the six-figure range. Mentor Precision and Synplicity were both about US$5K/yr last I checked. It's the ASIC tools that cost a hundred large per seat per year. And for your hundred large, you get the direct phone number of someone who can actually help you with your questions.

Now, as for the price of FPGA tools. Ever since I started doing FPGAs (XC3000 w/ViewLogic, back in the early 90s), I could never understand why the tools were so expensive ("expensive" defined as a couple thousand dollars per seat, plus FlexLM aggravation). After all, the tools were simply a means to an end: without using the tools, you can't design for the chips, so you're not going to buy any chips.

That exact argument had been raised on comp.arch.fpga easily over ten years ago. The usual response was, "for your two grand a year, you get our Xcellent support." If you think about it, charging some "nominal" fee aided the FPGA vendors by screening out the amateurs and others who'd chew up support hotline time. The software cost showed that you were "serious." Of course everyone knew that the Big Companies buying fab-runs of chips a month got the tools for free (and probably the direct phone number of a support person who knew his stuff).

But at some point, Brands A and X both realized that they'd sell more low-end chips in small quantities (making a larger profit, too) if they had a low-cost, or even free, version of the tools available. (I think this is called "listening to customers.") After all, not everyone is working on the latest/greatest chips. So they released "Web Pack" (what a stupid name) software, which supports most of the chips and has (by now) pretty much all of the extra stuff the full packages have. The caveat is that if you have problems, you have to file a "web case" and hope the support people can help you. If you're working with the newest/most-complicated chips (V4/V5 w/PPC core, etc) or if you want extra features (MicroBlaze core, ChipScope, etc) then you pay for it, which is reasonable. (I think this boils down to "don't charge for things the customer won't use.")

I don't do ASICs so I have no idea of how that EDA market works. I think, though, that there is a much larger requirement for correct-ness of results there, as an ASIC spin is remarkably expensive.

-a